mpack wrote:Both logs are truncated, either due to the abort or due to you grabbing the file while VirtualBox was still running.
I can assure you that vbox was NOT running when I copied the files. I've double-checked; these are the gzip'd logfiles as I saved them from the Log dialog box in the GUI. Perhaps this points to a part of the problem?
As a separate suggestion, why not just launch the VM headless, and RDP to it? That way you don't need the KVM switch.
I've done that at times. But there are also times when switching back and forth is necessary, such as when booting the other machine--it requires that the video card on that system syncs with the monitor. I do not seem to be able to do anything about that behavior.
But there are also other times, such as when RDP or SSH loses connectivity for whatever reason. The other box is a test box, and it is not unheard of for a test to fail, right?
(No, there were no failures on the test box at the time of the crashes on the first box. I switched back to monitor something periodically.)
Your problem may be caused because it's a USB hosted KVM: so each switch implies a set of USB devices being unplugged and replugged. VirtualBox needs to do nasty things when that happens, and it may not expect them to happen in parallel.
I agree completely. In fact, I highly suspect that is exactly the cause of the crashing. After all, once I switch back to USB 2 in the VM in question, the crashing no longer occurs.
What you say after that concerns me greatly, though. It is hard for me to hear that virtualbox cannot handle this switching without getting nasty about it. A VM is supposed to behave very close to actual hardware, so this is very disappointing. Stranger yet, this does NOT happen when running under USB 2.
I conclude that USB 3 requires more nastiness than USB 2 when performing such switching.
Does Oracle have plans to fully support USB 3 in the future?
Also... the switchbox is not even attached to the VM. The functionality should be handled by the host. I hope that the traffic generated by the switchbox would never reach virtualbox's ears.